We have created a dataset AIRPOSE used for aircraft pose estimation, which includes 200k synthetic images and 10k real images with annotations of bounding boxs and keypoints. 
There are in total 12 keypoints defined for aircrafts. The synthetic images were generated using 4 types of aircrafts(Boeing 787, 747; Airbus 320, 330) in total 15 different liveries under 18 different camera positions and 7 different climate conditions (clear skies, scattered cloud, broken cloud, over cast, rain, snow, storm) in 115 different airports with various lighting conditions. 
All of the 3D aircrafts are modeled realistically according to the real world aircrafts and each airport is also modeled based on the corresponding real-world airport. 
Both synthetic images and real images are set to the size of 1920x1080. 
We use MSFS2020 and Renderdoc for simulation and rendering analysis. 
The dataset is currently private for project purpose.